Generic rootless

Generic rootless

[Kensuke Matsuzaki]

[-- Attachment #1: Type: text/plain, Size: 438 bytes --]

Hello,

This uses miext/rootless. rootless mode create Windows window for each toplevel X 
window.
The remote X apps's reorder/maximize/minimize performance is improved.
And to click one X window doesn't raise all X window together.

Known problem:
 When using twm, I click window then that window raise top in Windows but it stay
previous position in X. I don't know how to avoid clicking Windows window raise
window.

Kensuke Matsuzaki

[-- Attachment #2: rootless.tar.bz2 --]
[-- Type: application/octet-stream, Size: 12425 bytes --]

Re: Generic rootless

[David Fraser]

Kensuke Matsuzaki wrote:

>Hello,
>
>This uses miext/rootless. rootless mode create Windows window for each toplevel X 
>window.
>The remote X apps's reorder/maximize/minimize performance is improved.
>And to click one X window doesn't raise all X window together.
>
>Known problem:
> When using twm, I click window then that window raise top in Windows but it stay
>previous position in X. I don't know how to avoid clicking Windows window raise
>window.
>
>Kensuke Matsuzaki
>  
>
Hi Kensuke

Wow, you keep on coming up with brilliant improvements

Thanks!
David

Re: Generic rootless

[Harold L Hunt II]

Kensuke,

Looks cool.  I really like the idea of doing this.  I had talked to 
Torrey about how to do this, but I never got around to it.

One problem I noticed right away:

Your diff for hw/xwin thinks it is creating the Imakefile as a new file 
because there was no xwin-20031017-1340/Imakefile file on your system. 
I think I can probably figure out what you were trying to do here 
(adding a few source and object files I assume).

Very impressive.

I won't have a chance to build this until a day or two from now.  I will 
give more feedback then.

Harold

Kensuke Matsuzaki wrote:

> Hello,
> 
> This uses miext/rootless. rootless mode create Windows window for each toplevel X 
> window.
> The remote X apps's reorder/maximize/minimize performance is improved.
> And to click one X window doesn't raise all X window together.
> 
> Known problem:
>  When using twm, I click window then that window raise top in Windows but it stay
> previous position in X. I don't know how to avoid clicking Windows window raise
> window.
> 
> Kensuke Matsuzaki

Re: Generic rootless

[Harold L Hunt II]

Kensuke,

I did a quick build test of the generic rootless code.  hw/xwin builds 
fine, but nothing is getting built in miext/rootless, nor has rootless 
extension been added to the XWin.exe link command (so there are 
unresolved references to rootless* symbols).  I am guessing that you had 
to modify Xserver/Imakefile or config/cf/cygwin.*, but forgot to send a 
diff for those.  Could you send that diff please?

Is the next step in this to pull the multi-window window manager back 
out of XWin.exe and make it a stand-alone executable, like XDarwin?  Or, 
is our current setup with the integrated multi-window wm the best solution?

Thanks for contributing,

Harold

Kensuke Matsuzaki wrote:

> Hello,
> 
> This uses miext/rootless. rootless mode create Windows window for each toplevel X 
> window.
> The remote X apps's reorder/maximize/minimize performance is improved.
> And to click one X window doesn't raise all X window together.
> 
> Known problem:
>  When using twm, I click window then that window raise top in Windows but it stay
> previous position in X. I don't know how to avoid clicking Windows window raise
> window.
> 
> Kensuke Matsuzaki

Re: Generic rootless

[Kensuke Matsuzaki]

[-- Attachment #1: Type: text/plain, Size: 1110 bytes --]

Harold,

> I did a quick build test of the generic rootless code.  hw/xwin builds 
> fine, but nothing is getting built in miext/rootless, nor has rootless 
> extension been added to the XWin.exe link command (so there are 
> unresolved references to rootless* symbols).  I am guessing that you had 
> to modify Xserver/Imakefile or config/cf/cygwin.*, but forgot to send a 
> diff for those.  Could you send that diff please?
Sorry. I forgot it.

> Is the next step in this to pull the multi-window window manager back 
> out of XWin.exe and make it a stand-alone executable, like XDarwin?  Or, 
> is our current setup with the integrated multi-window wm the best solution?
If possible, stand-alone executable seems better.
I read OroborOSX source some month ago. But I didn't know MacOS API so
I couldn't understand it. But I felt it is implemented Aqua like skin.
Their site say new v0.9 have many new features. I will read it.
I want someone who is familiar with XDarwin to comment.


By the way, Greg Parker's xroot.c is very useful when I use rootless mode.
http://sealiesoftware.com/

Kensuke Matsuzaki

[-- Attachment #2: Imakefile.patch --]
[-- Type: application/octet-stream, Size: 1472 bytes --]

Index: Imakefile
===================================================================
RCS file: /cvsroot/xoncygwin/xc/programs/Xserver/Imakefile,v
retrieving revision 1.3
diff -u -r1.3 Imakefile
--- Imakefile	20 Sep 2003 05:49:12 -0000	1.3
+++ Imakefile	20 Oct 2003 03:56:29 -0000
@@ -1209,19 +1209,21 @@
 #else
 LAYERDIR = 
 #endif
+ROOTLESSDIR = miext/rootless
 XWINDDXDIR = hw/xwin
 XWINPARSERDIR = hw/xfree86/parser
 XWINPARSERLIB = $(XWINPARSERDIR)/LibraryTargetName(xf86config)
+ROOTLESSLIB = $(ROOTLESSDIR)/LibraryTargetName(rootless)
 XWINLIB = $(XWINDDXDIR)/LibraryTargetName(Xwin)
 XWINDIRS = $(STDDIRS) $(FBDIR) $(SHADOWDIR) $(LAYERDIR) $(XWINDDXDIR) \
-	   $(DEPDIRS) $(XWINPARSERDIR)
+	   $(DEPDIRS) $(XWINPARSERDIR) $(ROOTLESSDIR)
 XWINOBJS = $(XWINDDXDIR)/stubs.o $(XWINDDXDIR)/XWin.res
 #if 0
 XWINLIBS = PreFbLibs $(XWINLIB) FbPostFbLibs $(LAYER) $(SHADOW) \
            $(XWINPARSERLIB)
 #else
 XWINLIBS = PreFbLibs $(XWINLIB) FbPostFbLibs $(SHADOW) \
-           $(XWINPARSERLIB)
+           $(XWINPARSERLIB) $(ROOTLESSLIB)
 #endif
 XWINSYSLIBS = $(FONTLIBS) $(LDPRELIBS) $(XLIB) $(SYSLIBS) -lgdi32
 
@@ -1352,7 +1354,7 @@
 
 
 CFBDIRS = $(CFB8DIR) $(CFB16DIR) $(CFB24DIR) $(CFB32DIR)
-MIEXTDIRS = $(SHADOWDIR) $(LAYERDIR)
+MIEXTDIRS = $(SHADOWDIR) $(LAYERDIR) $(ROOTLESSDIR)
 IPLANDIRS = $(IPLAN2P2DIR) $(IPLAN2P4DIR) $(IPLAN2P8DIR)
 DDXDIRS = $(DECWSDDXDIR) $(SUNDDXDIR) $(LYNXDDXDIR) \
 	  $(HPDDXDIR) $(XFREE86DDXDIR) $(XWINDDXDIR) $(DARWINDDXDIR) \

Re: Generic rootless

[Harold L Hunt II]

Kensuke,

Okay, I got it to build and run.  It is very nice.

What you think about renaming old '-rootless'to '-oldrootless' and 
calling the new '-win32rootless' just '-rootess'?

-win32rootless is going to cause a lot of trouble for people to 
remember, and most people will get confused about the distinction 
between -rootless and -win32rootless.  Is there any reason to suspect 
that the old '-oldrootless' will remain in use after the new '-rootless' 
is available?  It would seem to me that we might even want to just 
replace the original rootless functionality with the new functionality 
that uses the shared code.

Anybody have some input on this?

Harold

Kensuke Matsuzaki wrote:

> Harold,
> 
> 
>>I did a quick build test of the generic rootless code.  hw/xwin builds 
>>fine, but nothing is getting built in miext/rootless, nor has rootless 
>>extension been added to the XWin.exe link command (so there are 
>>unresolved references to rootless* symbols).  I am guessing that you had 
>>to modify Xserver/Imakefile or config/cf/cygwin.*, but forgot to send a 
>>diff for those.  Could you send that diff please?
> 
> Sorry. I forgot it.
> 
> 
>>Is the next step in this to pull the multi-window window manager back 
>>out of XWin.exe and make it a stand-alone executable, like XDarwin?  Or, 
>>is our current setup with the integrated multi-window wm the best solution?
> 
> If possible, stand-alone executable seems better.
> I read OroborOSX source some month ago. But I didn't know MacOS API so
> I couldn't understand it. But I felt it is implemented Aqua like skin.
> Their site say new v0.9 have many new features. I will read it.
> I want someone who is familiar with XDarwin to comment.
> 
> 
> By the way, Greg Parker's xroot.c is very useful when I use rootless mode.
> http://sealiesoftware.com/
> 
> Kensuke Matsuzaki

Re: Generic rootless

[Alan Hourihane]

On Mon, Oct 20, 2003 at 01:27:59PM -0400, Harold L Hunt II wrote:
> Kensuke,
> 
> Okay, I got it to build and run.  It is very nice.
> 
> What you think about renaming old '-rootless'to '-oldrootless' and 
> calling the new '-win32rootless' just '-rootess'?
> 
> -win32rootless is going to cause a lot of trouble for people to 
> remember, and most people will get confused about the distinction 
> between -rootless and -win32rootless.  Is there any reason to suspect 
> that the old '-oldrootless' will remain in use after the new '-rootless' 
> is available?  It would seem to me that we might even want to just 
> replace the original rootless functionality with the new functionality 
> that uses the shared code.
> 
> Anybody have some input on this?

I agree Harold.

Deprecate the old one and remove the code, unless there's some
benefits to using it.

Alan.

Re: Generic rootless

[Brian E. Gallew]

Harold L Hunt II wrote:
> -win32rootless is going to cause a lot of trouble for people to 
> remember, and most people will get confused about the distinction 
> between -rootless and -win32rootless.  Is there any reason to suspect 
> that the old '-oldrootless' will remain in use after the new '-rootless' 
> is available?  It would seem to me that we might even want to just 
> replace the original rootless functionality with the new functionality 
> that uses the shared code.
> 
> Anybody have some input on this?

Post a binary and I'll let you know!  8-}

I think the real question here is: how is the new rootless mode going to 
affect multi-window mode (if at all)?  I've used the old rootless mode, 
and like it a lot, BUT it's really convenient to have my windows show up 
in the task bar/menu.

Re: Generic rootless

[Harold L Hunt II]

Brian E. Gallew wrote:

> Harold L Hunt II wrote:
> 
>> -win32rootless is going to cause a lot of trouble for people to 
>> remember, and most people will get confused about the distinction 
>> between -rootless and -win32rootless.  Is there any reason to suspect 
>> that the old '-oldrootless' will remain in use after the new 
>> '-rootless' is available?  It would seem to me that we might even want 
>> to just replace the original rootless functionality with the new 
>> functionality that uses the shared code.
>>
>> Anybody have some input on this?
> 
> 
> Post a binary and I'll let you know!  8-}

Okay, unfortunately this can only be built from CVS head, so it can't be 
distributed as a 'test' release via setup.exe.  Ugh... this is going to 
get me into the same sort of thing that was the Server Test Series.  I 
really don't want to go down that route again... but I will post one 
binary for people to check out:

http://www.msu.edu/~huntharo/xwin/devel/server/XWin-rootless-20031020-1341.exe.bz2 
(1.3 MiB)

If you don't know what to do with the above, then please just forget 
about it; do not ask for help, you will not receive any.  This is only 
to enable other developers and very advanced users to provide feedback 
on the new shared rootless code.

> I think the real question here is: how is the new rootless mode going to 
> affect multi-window mode (if at all)?  I've used the old rootless mode, 
> and like it a lot, BUT it's really convenient to have my windows show up 
> in the task bar/menu.

This won't have any effect on multi-window mode for now.  The rootless 
mode still requires an external window manager.  We may rewrite the 
current multi-window mode as an external window manager, which would 
allow the new rootless mode and the new external window manager to do 
the same thing as the current multi-window mode.

Harold

Re: Generic rootless

[Harold L Hunt II]

Alan Hourihane wrote:

> On Mon, Oct 20, 2003 at 01:27:59PM -0400, Harold L Hunt II wrote:
> 
>>Kensuke,
>>
>>Okay, I got it to build and run.  It is very nice.
>>
>>What you think about renaming old '-rootless'to '-oldrootless' and 
>>calling the new '-win32rootless' just '-rootess'?
>>
>>-win32rootless is going to cause a lot of trouble for people to 
>>remember, and most people will get confused about the distinction 
>>between -rootless and -win32rootless.  Is there any reason to suspect 
>>that the old '-oldrootless' will remain in use after the new '-rootless' 
>>is available?  It would seem to me that we might even want to just 
>>replace the original rootless functionality with the new functionality 
>>that uses the shared code.
>>
>>Anybody have some input on this?
> 
> 
> I agree Harold.
> 
> Deprecate the old one and remove the code, unless there's some
> benefits to using it.

Alan,

Input noted.  We agree completely.

Harold

Re: Generic rootless

[Brian E. Gallew]

Harold L Hunt II wrote:
> Okay, unfortunately this can only be built from CVS head, so it can't be 
> distributed as a 'test' release via setup.exe.  Ugh... this is going to 
> get me into the same sort of thing that was the Server Test Series.  I 
> really don't want to go down that route again... but I will post one 
> binary for people to check out:

Works like a charm.  My opinion: if you like the code, then replace the 
old version.

Re: Generic rootless

[Kensuke Matsuzaki]

Harold,

> Okay, I got it to build and run.  It is very nice.

Good.


> What you think about renaming old '-rootless'to '-oldrootless' and 
> calling the new '-win32rootless' just '-rootess'?

I think that's good, and acutually I wont to do so.

Kensuke Matsuzaki

[PATCH] Generic rootless

[Kensuke Matsuzaki]

[-- Attachment #1: Type: text/plain, Size: 183 bytes --]

Harold,

The old rootless mode is replaced new rootless mode which use generic
rootless library.
And a problem that redrawing fails after window resizing is fixed.

Kensuke Matsuzaki

[-- Attachment #2: win32rootless-20031022-1414.patch.bz2 --]
[-- Type: application/octet-stream, Size: 11270 bytes --]

Re: [PATCH] Generic rootless

[Harold L Hunt II]

Kensuke,

I just wanted to drop you a line to let you know that I have received 
this patch and I will be looking into it... I have just been busy with 
other things for a while.

By the way, I am looking into ways to allow you and other developers to 
have direct CVS access to the main source tree (if that is what you can 
call the discussion on devel@xfree86.org).  The intention here is that I 
will no longer slow you down when this is finished :)

Harold

Kensuke Matsuzaki wrote:

> Harold,
> 
> The old rootless mode is replaced new rootless mode which use generic
> rootless library.
> And a problem that redrawing fails after window resizing is fixed.
> 
> Kensuke Matsuzaki